0
\$\begingroup\$

History/context: I talk about a 15Mhz noise when drill machine or any kind of machine activation in my network, before, at other post. By the moment neither RC, RL or RLC butterworth low pass filter designed for 15Khz is not having any effect at PCB voltage inputs line: no attenuation, no wave form change, etc.. the same that being a wire instead of a circuit. This part was treated there. About system power. The noise was also appearing when using dual power supply, for laboratory usage. It was bought at Farnell or RS distributors. The other power bank supply is being used for years inside banks that were manufactured here.

The noise appears not only at Vin, Vout circuit lines. It has also seen at VCC pins from connectors of other old PCBs (12V, 24V, 5V), even at Raspberry pi, Arduino, etc GPIO and/or power lines. But they seem not to have behavior problems as my PCB. That is the thing why I try a second why to face the problem. It seems that any tested hardware could avoid this noise.

The problem to solve: My functional problem when noise appears is located at reset line. It resets devices configuration: configuration register value became a 0x00 and devices loses their configuration. When talk about devices I mean MCP23S17.

Reset line was first implemented from Raspberry Pi GPIO output. After the problem of noise, we decided to force devices reset pin to 3.3V. So it will always be enabled. Now, there is no line between the Raspb GPIO output and devices reset pin. Trace was removed. They always are enabled when the PCB is powered. One board has 8 devices, but I have other one that has only one of them. Two both resets are forced to VCC = 3.3V.

Focus: I would like to focus the post to the problematic effect on my PCB and ask for other suggestions in the reset way: try to solve the noise effects on the reset pin. As by the moment the only problem that the transcient is causing to the system is the undesired reset, could I try to avoid the transcient effect at reset line?

The noise causes that the reset signal falls down from 3.3V to 0V for a while and resets the chip. 3.3V is a 5V regulator output. 5V comes from a commercial source supply, a power bank. But noise can be seen at 5V regulator input and 3.3V regulator output, it also can be seen at reset pin of the chip.

Drill machine noise at 5v and 3.3v RST

Blue is 5V (bank Vout = regulator Vin) and yellow is reset pin (tied to 3.3V, so regulator Vout). As can it be seen there when drill machine is on noise transient appears for a while. I measured the freq and it is always about 15Mhz.

My ideas for solving it: So I thought about to add some circuitry for avoiding the fallen voltage (from 3.3V to 0V) at chip reset pin. It must still being 3.3V although this noise. Looking for reset circuits I have found this configuration and I tried adding it:

level shifter

I get experimentaly the best adjustment at:

  • R = 20K
  • C= 120pF
  • Q = BC567
  • "reset" wire is forced to 5V (bank power Vout)

I put a probe at 3.3V and the other one to the Emiter node (that will connect reset device pin).

I got:

Reset noise attenuation

  • Blue is emiter probe
  • Yellow is 5V

This kind of circuit adapter seems to filter and to reduce noise effects. But this is my best result. By the moment I don't know it this level of attenuation will be enough when I connect it to the real reset. The next step is to connect it to the system, set up, check it and take measures. It seems to be a good way. I don't get this noise level before.

Askings: Can anyone suggest:

  • any modification of the circuit above in order to improve a bit more my wave form results?
  • other kind of circuitry for avoiding pulses at reset pin?

Please if someone has better ideas let me now. I will be grateful.

\$\endgroup\$
5
  • 3
    \$\begingroup\$ "By the moment no pass low filter at pcbs inputs could have any effect." - what have you tried? "5V comes from a commercial source supply, a power banck." - Try powering your board with a battery, do you still get noise on the +3.3V and reset line? What external I/O signals are connected to your board, and where do they come from / go to? Does any noise appear on them? \$\endgroup\$ Commented Dec 3, 2020 at 17:52
  • 4
    \$\begingroup\$ Suvi_Eu - Hi, Some parts of your story are difficult to understand due to translation issues. As well as performing the suggested tests from Bruce Abbott, I recommend that you edit the question and (a) add oscilloscope trace images of the "noise" on the 5V i/p, 3.3V regulator o/p and -Reset signal; (b) explain why are trying to fix -Reset "high" instead of solving the drop of the power rail; (c) add a schematic; (d) add some photos of the actual hardware; (e) add some history e.g. is this a new design? Has that power source (power bank) been used for other devices successfully? etc. Thanks \$\endgroup\$
    – SamGibson
    Commented Dec 3, 2020 at 23:55
  • \$\begingroup\$ I will edit the post right now. I will try to improve the explanation if possible. \$\endgroup\$
    – Suvi_Eu
    Commented Dec 4, 2020 at 7:40
  • 1
    \$\begingroup\$ Please share schematic and layout. How's you ground plane, decoupling and layout situation? \$\endgroup\$
    – winny
    Commented Dec 4, 2020 at 10:42
  • \$\begingroup\$ @Suvi_Eu - Hi, Thanks for adding some details. However I didn't see any updates that I could understand for my earlier points (b), (c), (d) or (e). So I plan to stop after this comment. Again, I will try to say clearly, that aiming to fix the -Reset signal is the wrong thing to do IMHO. That is a symptom, not a cause! The 'scope traces show such massive transients on the 5V rail (and therefore 3.3V rail), that of course, if you tie -Reset directly to the 3.3V rail, then -Reset will experience the same transient! Focus on preventing the power rail transients instead. Good luck! \$\endgroup\$
    – SamGibson
    Commented Dec 4, 2020 at 21:29

1 Answer 1

0
\$\begingroup\$

1.first of all you need isolate the ac part from dc part of your circuit. note that you can use some element such as optocupler 2.use shiled cable to your device if needed. 3.apply the polygon pour to your pcb. 4.add a metal surface which connected to the ground, under your dc part. 5.if you have any relay in your circuit, use snaber on it.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.